India Puts OpenAI and Anthropic Cybersecurity Plans on Hold for Now
- pradeep
- 2 days ago
- 2 min read
A department under the Ministry of Electronics and Information Technology (MeitY) has reportedly advised Union ministries not to deploy cybersecurity models from OpenAI and Anthropic for the time being.
According to an exclusive report by ThePrint, an office memorandum was circulated after representatives of the two AI companies approached several ministries with proposals to use their models for cybersecurity and related government functions. The reported direction is described as a temporary pause rather than a permanent ban. (ThePrint)
Why Has Deployment Been Paused?
The MeitY department reportedly believes that deploying such systems at this stage would be premature.
A source cited in the report said the objection concerns the timing and readiness of deployment, rather than a complete rejection of OpenAI or Anthropic technology. (ThePrint)
The report did not confirm how many ministries were approached or the seniority of the officials who attended the meetings.
Finance Ministry Explored GPT-5.5 for Cybersecurity
The Ministry of Finance was reportedly among the departments that sought guidance on using advanced AI systems.
According to the report, the ministry submitted a six-page proposal examining agentic AI, AI-powered vulnerability discovery and the potential deployment of OpenAI’s GPT-5.5 for cybersecurity work. The MeitY department reportedly declined the request in a memorandum issued last week. (ThePrint)
What Can These AI Models Do?
Frontier AI models are increasingly marketed for defensive cybersecurity functions, including:
Identifying security weaknesses in software
Reviewing computer code
Assisting with incident investigation and triage
Carrying out multi-step cybersecurity tasks with limited supervision
However, these capabilities are considered dual-use. A system capable of finding vulnerabilities for defensive purposes could potentially identify the same weaknesses for malicious exploitation. (ThePrint)
Data Security and Foreign Infrastructure Concerns
The reported advisory comes amid unresolved questions about how foreign AI models should be used with government information.
Current government rules address areas such as data localisation, approved cloud infrastructure and the handling of sensitive information. However, the report says they do not fully clarify issues such as where AI processing occurs, what information vendors retain or whether model infrastructure is located outside India. (ThePrint)
These concerns become especially important when AI is used in cybersecurity, critical infrastructure or other strategic government functions.
India’s Sovereign AI Push
The development also comes as India promotes a sovereign AI ecosystem aimed at reducing dependence on foreign technology for strategically important applications.
The IndiaAI Mission, approved in 2024 with an allocation of approximately ₹10,372 crore, supports computing infrastructure, datasets and Indian foundation-model developers, including Sarvam AI and the BharatGen consortium. (ThePrint)
Cybersecurity is considered a particularly sensitive area because CERT-In operates under MeitY, while critical-information infrastructure protection is coordinated through the National Critical Information Infrastructure Protection Centre.
Not a Permanent Ban
The reported memorandum does not appear to prohibit ministries from ever using OpenAI or Anthropic models.
Instead, it signals that the government may want clearer procurement rules, security safeguards, data controls and technical evaluations before allowing foreign frontier AI models to be deployed in sensitive cybersecurity operations. This is an inference based on the concerns and policy context described in the report. (ThePrint)
Neither MeitY, OpenAI nor Anthropic was quoted in the report as issuing a formal public response to the reported memorandum.
Comments